Abstract: A potentially valuable tool in medical imaging is the development, and integration with PACS, of systems which enhance the interpretive accuracy of the user - his ability, given a set of findings (in the broad sense, including clinical information about the patient as well as characteristics of the lesion being analyzed), to assign the proper disease label, or diagnosis, to them. Such systems, which we call here interpretive tools (IT), contain a variety of types of information about diseases and their radiologic diagnosis. They can contain information about large numbers of diseases, including statistical information (incidence, characteristic anatomical locations, association with age and gender or with other diseases, probabilities of various findings given a disease), textual information (description of diseases, treatment, literature references, lists of other entities that might be confused with the disease of interest, additional diagnostic points that may not be represented, or even representable, within IT), and image-based information (typical and atypical examples for each entity and radiographic finding, examples of normal anatomy). These databases can be used both for teaching purposes and as a tool for improving interpretive accuracy $LB@Swett, 1987$RB@. This paper describes some of the requirements for these databases and then discusses early work on the implementation of DDx, an IT whose domain is neuroradiology. !6
展开▼
